In-vivo imaging of neural activity with dynamic vision sensors

Abstract

Optical recording of neural activity using calcium or voltage indicators requires cameras capable of detecting small temporal contrast in light intensity with sample rates of 10 Hz to 1 kHz. Large pixel scientific CMOS image sensors (sCMOS) are typically used due to their high resolution, high frame rate, and low noise. However, using such sensors for long-term recording is challenging due to their high data rates of up to 1 Gb/s. Here we studied the use of dynamic vision sensor (DVS) event cameras for neural recording. DVS have high dynamic range and a sparse asynchronous output consisting of brightness change events. Using DVS for neural recording could avoid transferring and storing redundant information. We compared the use of a Hamamatsu Orca V2 sCMOS with two advanced DVS sensors (a higher temporal contrast sensitivity 188×180 pixel SDAVIS and a 346×260 pixel higher light sensitivity backside illuminated BSIDAVIS) for neural activity recordings with fluorescent calcium indicators both in brain slices and awake mice. The DVS activity responds to the fast dynamics of neural activity, indicating that a sensor combining SDAVIS and BSIDAVIS technologies would be beneficial for longterm in-vivo neural recording using calcium indicators as well as potentially faster voltage indicators.

Live Demonstration: Front and Back Illuminated Dynamic and Active Pixel Vision Sensors Comparison

“…The latter induces more "leak events" and thus more noise. Overall, the BSI DAVIS is 4 times more sensitive than the FSI DAVIS, showing possible advantages in low light applications such as shown in previous works on neural imaging of [2] [3].…”

Event-Based Sensing and Signal Processing in the Visual, Auditory, and Olfactory Domain: A Review

“…Using CMOS sensors is very challenging as they require high data rates of up to 1 Gb/s. To overcome this, a DVS camera is used for the task in Taverni et al (2017), that suggests long-term use of the sensor in neural recordings can be very beneficial.…”

Front and Back Illuminated Dynamic and Active Pixel Vision Sensors Comparison

“…CONCLUSION Compared to the previous generation DAVIS240C, the FSI DAVIS has about 3.5X higher QE and the BSI DAVIS has 13X higher QE. The BSI DAVIS, with its 4X higher QE than the FSI DAVIS, has clear advantages for low light applications such as the fluorescent calcium neural imaging of [4], where the lighting spectrum can be controlled and where the lower MTF can be tolerated. However, the increased sensitivity of leak events to parasitic photocurrent caused by NIR currently limits applications of the BSI DAVIS to controlled lighting.…”

“…The growing interest in DAVIS and DVS sensors has pushed the exploration of new techniques in order to reach higher performance in terms of sensitivity and low illuminance operation, so to extend applications in new fields such as fluorescent microscopy [4] and astronomy [5].…”